What is the Best Way to Do Something? A Discreet Tour of Discrete Optimization

In mathematical optimization, we want to find the best possible solution for a decision-making problem. Curiously, these problems are harder to solve if they have discrete decisions. Imagine that you would like to buy chocolate: you can buy no chocolate or one chocolate bar, but typically you cannot buy just half of a bar. Now imagine that you could also buy many other items, and that you need to meet nutritional needs while minimizing the grocery bill. With more options and more demands, finding the best solution becomes trickier. But since many real-world settings benefit from mathematical optimization, such as scheduling trains and flights, planning truck deliveries, and making better investment decisions, these problems are widely studied in a branch of mathematics called Operations Research (OR). Sometimes we can simply write the mathematical model and find an optimal solution with OR software, but for larger problems we may need to develop new mathematical models and even write our own algorithms. We explore both cases with a simple and well-known problem (the traveling salesperson problem), some computer programming (in Python), and software that is free for academic use (Gurobi). All the code and data used is available at: https://github.com/thserra/discreet

💡 Research Summary

The paper serves as a pedagogical introduction to discrete optimization, aimed at undergraduate students with basic knowledge of algorithms, programming (preferably Python), calculus, and linear algebra. It begins by motivating the need for optimization through everyday examples—first a simple fruit‑buying problem, then the classic traveling salesperson problem (TSP). The author uses the fruit example to illustrate core concepts: decision variables, objective functions, constraints, and the importance of explicitly stating variable domains (continuous vs. integer). By modeling the purchase of apples and oranges with costs and potassium requirements, the paper shows that the continuous relaxation yields a cheaper fractional solution (≈1.013 oranges) while the integer model forces whole fruits, leading to a higher cost (two apples). This contrast underscores why discrete decisions make problems harder and why integer programming is essential for realistic decision‑making.

Building on this, the author introduces a more complex integer linear programming (ILP) model involving four fruits (apples, oranges, bananas, pears) and three nutritional constraints (potassium, calcium, fiber). The model is presented in standard ILP form: minimize (c^T x) subject to (A x \ge b), bounds (l \le x \le u), and integrality (x \in \mathbb{Z}^n). The paper explains the notation for cost vectors, coefficient matrices, and bounds, and discusses binary variables as a special case of integer variables.

A substantial portion of the text is devoted to the relationship between ILP, linear programming (LP) relaxations, and mixed‑integer linear programming (MILP). The LP relaxation is obtained by allowing all variables to be continuous, turning the problem into a potentially easier one whose feasible region contains that of the ILP. The author emphasizes that sometimes the LP optimal solution is already integral, which can be exploited computationally. MILP is introduced as a hybrid where a subset of variables remains integer while the rest are continuous, formalized as (x \in \mathbb{Z}^p \times \mathbb{R}^{n-p}). The discussion includes the practical significance of these variations and how they affect solver performance.

To reinforce learning, the paper intersperses four exercises. Exercise 1 asks whether the continuous optimal value is always lower than the integer one. Exercise 2 challenges the reader to adjust the potassium requirement so that both models share the same optimal solution, and to minimally change orange prices to make two oranges optimal in the integer model. Exercise 3 extends the model to handle existing inventory, differing buy‑sell prices, and the impact on objective and constraints. Exercise 4 presents a knapsack‑type problem using chocolate bar happiness scores and prices, illustrating another classic discrete optimization scenario.

Implementation details are provided for solving the models with Python and the free academic license of Gurobi. All source code and data are hosted on GitHub (https://github.com/thserra/discreet), enabling readers to replicate experiments, modify models, and explore solution behavior. The author also discusses practical modeling pitfalls, such as confusing problem description with mathematical formulation, and the importance of clearly distinguishing between a problem and its various possible models.

Overall, the paper follows a clear workflow: problem definition → mathematical modeling (variables, objective, constraints) → LP relaxation → integer/MILP solution → result interpretation → model refinement. By combining intuitive examples, rigorous formulation, hands‑on coding, and reflective exercises, it equips beginners with both the theoretical foundation and practical skills needed to tackle real‑world discrete optimization problems.